Randomised controlled trials of complex interventions

Abstract

Although all randomised controlled trials (RCT) share much in common in terms of design, conduct, analysis and reporting, trials of complex interventions (loosely " interventions with several interacting components ") can have special issues which need appropriate consideration to deliver high quality and credible evidence to change or inform practice. For example, blinding the intervention may be impos-sible (a surgical compared with a medical intervention), and standardising the intervention (as would be obliga-tory and easy in a drug trial) difficult to define and harder still to achieve (for example, reaching a therapeu-tic alliance between patient and therapist in a cognitive behavioural therapy RCT). In addition, often complex intervention trials have patient reported primary out-comes (for example, quality of life) that present chal-lenges (e.g. deaths, missing data). Safety reporting in complex intervention trials often presents particular challenges, struggling to adapt methodologies designed and developed for regulatory drug trials. Designers of complex intervention trials may be forced to make assumptions about critical features of the inter-vention (e.g. intensity, and duration) and how to mea-sure the treatment effect without the funds or the time to develop these aspects in a similar manner to the Phase I-IV development of drug trials. Frequently com-plex intervention trials are underpowered to detect important effects due to their expense and perceived dif-ficulties. Often times one is left wondering whether the failure of a complex intervention trial to detect a treat-ment effect was down to sub optimal design, rather than strong & robust evidence that it does not work. This talk will highlight various methodological chal-lenges in the various stages of the design and running of complex intervention trials, including the development of evidence based interventions and outcomes, through to interpretation of often difficult analyses and convin-cing skeptical peers and journals of the robustness of findings. The talk will be illustrated by practical exam-ples from many complex intervention trials the speaker has been involved in, including several large, multicen-tre, pragmatic trials funded by the NIHR/HTA, and more explanatory trials funded by MRC/EME. It will also draw on first-hand experience of sitting as an inde-pendent statistician / methodologist / trialist on various Trial Steering Committees and Data Monitoring Boards, as well as on various funding panels and journals.